Dashboard Information:

In my ILE project I am trying to use latent trajectory factor analysis (LTFA) to infer from cognitive test performances the longitudinal change of cognitive ability among Alzheimer’s Disease patients, as well as its relationship between those cognitive test performances.

This dashboard summarizes the Health Brain Study data that I use and model outcomes from my longitudinal cognitive change inference. It includes exploratory visualizations for missingness of longitudinal visits of the dataset, summarize of longitudinal trends by visits for cognitive tests, and finally, the estimated Cognitive ability’s relationship with cognitive test performance.

Github link:https://github.com/Michael89757/DATA_555_Dashboard

Project Impact:

Impact: This dashboard helps understand the Emory Health Barin Study data with a interactive way, and it also visualize longitudinal cognitive trajectories and its connection with cognitive tests performances, can help to identify patterns that may indicate early neurodegenerative changes and testify efficiency of cognitive tests in pre-clinical measurements.


Dataset Description

The dataset used in this analysis originates from the Emory Health Brain Study (EHBS). It is a sub-study within the Emory Healthy Aging Study (EHAS), which is an observational study designed to advance the understanding of healthy aging and the development of agerelated diseases through well-defined, community-based prospective cohorts.The derived dataset, cleaned collaboratively, contains individual-level identifiers (GUID), record IDs for each visit, visit event names, demographic information of the participants, and their performance on various cognitive tests across visits.EHBS specifically recruits up to 2,500 EHAS participants aged 50 to 75 who have no diagnosis of AD.

The following table lists the names of cognitive tests and their corresponding meanings:

Cognitive Test Names and Corresponding Meanings
Cognitive.Test.Name Meaning
RAVLT A6 Rey Auditory Verbal Learning Test A6 Total Correct (Immediate Recall)
RAVLT A7 Rey Auditory Verbal Learning Test A7 Total Correct (Delayed Recall)
JOLO Judgment of Line Orientation Total Number Correct
MiNT The Multilingual Naming Test Total Correct (sum of uncued + semantic cue, max = 32)
SDMT Symbol Digit Modalities Test Total Correct
NumSpanBack Number Span Backward Test: Longest Span Correct
NumSpanFwd Number Span Forward Test: Longest Span Correct
Trails_A Trail Making Test (Part A)
Trails_B Trail Making Test (Part B)
Verbal Fluency F Verbal Fluency - Number of F-words generated in 60 seconds
Verbal Fluency L Verbal Fluency - Number of L-words generated in 60 seconds
RCFT_Immediate Rey Complex Figure Test (Immediate Recall)
RCFT_Delayed Rey Complex Figure Test (Delayed Recall)
Category Total Category Fluency - Total Animals

Widget 1: Summary of Cognitive Test Observations by Each Visit

Column

Column

This plot illustrates record availability per visit for each cognitive test and provides a visualization of drop out status.

Key Takeaways:

  • SDMT and Verbal Fluency L have notably fewer records at enrollment stage hence is excluded from out model. Consider excluding sparse tests from the final model to reduce noise.
  • There is significant amout of drop out after Baseline visit.
  • Use this summary to guide which cognitive measures are well-represented longitudinally.

Widget 2: Standardized Cognitive Scores by Visit

Column

Column

This plot illustrates distribution of cognitive test perfromances across each visits to visualize test performance longitudinal change. By clicking on the cognitive test on legends user can filter test to be presented

Key Takeaways:

  • Most scores seemingly have a steady distribution for medians with slightly decrease after visit 3
  • The spread varies for different visit, probably due to variant sample size for each visit.
  • The outliers are evident for most of the tests indicating significant difference among patient test performacnes.

Widget 3: LTFA Model-Implied Cognitive Test Scores

Column

Column

This plot illustrates the distribution of cognitive test across estimated latent factor (cognitive ability). By clicking on the Event Name user can see the distribution of cognitive test scores across different cognitive ability for each visit.

Key Takeaways:

  • RAVLT A6 & A7 Total Correct, RCFT(Immediate Recall) and Category Fluency Total Animals have clear distinguishing between low and high cognitive ability in performances indicated by its slope and the distinguishing is evident with the passage of time.
  • All cognitive tests seems have a overall better performance each time (higher intersection) indicating possible practice effects.
  • Users can use this output to guide justification and selection of cognitive tests.